I’m opening NEES Core Engine for developer feedback.
NEES Core Engine is a governed AI runtime layer that sits between an AI application and the model provider.
User → App → NEES Core Engine → Model Provider → Governed Response
The goal is not to replace the model.
The goal is to make AI product behavior more controllable, traceable, and reviewable.
Why I’m building this
Most AI product failures are not only model failures.
In real AI workflows, the failure often comes from the system around the model:
unclear role boundaries
messy memory/context scope
missing escalation paths
weak permission boundaries
no traceability
no reviewable decision history
behavior drift across sessions
prompt fixes scattered across the product
A model can pass a safety filter and still behave incorrectly for the product.
That is the gap I’m trying to explore with NEES Core Engine.
What NEES Core Engine focuses on
NEES Core Engine is designed around runtime governance for AI products:
behavior governance
role consistency
memory boundaries
intent-aware policy decisions
runtime trace IDs
escalation/fallback visibility
reviewable AI responses
Who I’m looking for
I’m looking for developers building or testing:
AI agents
customer support bots
internal copilots
workflow automation tools
AI apps using memory
AI apps using tools/actions
products where role, tone, escalation, or traceability matter
Developer preview
GitHub repo:
https://github.com/NEES-Anna/nees-core-developer-preview
Live sample app:
The preview includes:
Python quickstart
Node.js quickstart
cURL example
API reference
governance flow docs
API key request template
developer feedback template
What feedback would help most
If you test it, I’d love feedback on:
Is the quickstart clear?
Does the governance flow make sense?
Are trace IDs useful?
Is the response metadata helpful?
What fields would you need in production?
Where does the runtime feel incomplete?
What failure modes should NEES handle better?
What would stop you from integrating this into a real workflow?
This is an early developer preview, so I’m not looking for praise.
I’m looking for honest technical feedback from builders.
Even 15 minutes of testing would help shape the next version of NEES Core Engine.
Top comments (1)
I am not an ai expert, just a curious user!
I tried the live app and its really good!
I like the "Governance".I just want to know how it works?
Do you train it on specific keywords like "unethical"?